Background
The recently published INTERGROWTH-21st Project international population standard for newborn size is intended for global use, but its ability to identify small infants at risk of adverse outcomes in a general obstetric population has not been reported.
Objective
The objective of the study was to compare adverse neonatal outcomes among small-for-gestational-age (SGA) infants between the INTERGROWTH-21st standard and a customized birthweight standard (accounting for maternal characteristics of height, weight, parity, and ethnicity). We hypothesized that in a multiethnic general obstetric population in Auckland, New Zealand, a customized birthweight standard would better identify SGA infants at-risk of neonatal morbidity/mortality and stillbirth than the INTERGROWTH-21st standard.
Study Design
Using prospectively gathered maternity data from a general obstetric population in Auckland, New Zealand, from 2006 to 2013 (n = 53,484 births at ≥ 33 weeks), infants were classified as SGA (birthweight < 10th centile) by INTERGROWTH-21st and customized standards. Infants were further categorized as SGA by both criteria, INTERGROWTH-21st only, customized only, or not SGA (met neither criteria). Composite adverse neonatal outcome was defined as neonatal death, neonatal intensive care admission > 48 hours, or ventilation > 4 hours or 5-minute Apgar score < 7. Relative risks for primary outcomes were estimated using modified Poisson regression, with the non-SGA group as the referent.
Results
Incidence of SGA was 4.5% by INTERGROWTH-21st and 11.6% by customized standard. Compared with those not SGA, infants identified as small for gestational age by both criteria had the highest risk of adverse neonatal outcome (relative risk [RR], 4.1, 95% confidence interval [CI], 3.7–4.6) and stillbirth (RR, 8.3, 95% CI, 5.1–13.4). Infants SGA by customized standard only (n = 4015) had an increased risk of adverse neonatal outcome (RR, 2.0, 95% CI, 1.8–2.2) and stillbirth (RR, 3.0, 95% CI, 1.7–5.3). Few infants were identified as SGA by INTERGROWTH-21st only (n = 172), and risks of adverse neonatal outcome and stillbirth were not increased. Findings were unchanged when analyses were limited to term infants (n = 50,739). The INTERGROWTH-21st standard identified more Indian (12.8%) and Asian (5.8%) but fewer European (3.0%) and Pacific (2.9%) infants as SGA ( P < .01). Customized criteria identified more than 3 times as many SGA infants among Maori (14.5%), Pacific (13.5%), and European (11.2%) infants and twice as many among Asian (10.3%) infants (P<0.01) compared with INTERGROWTH-21st criteria. The majority of SGA infants by INTERGROWTH-21st only were born to Indian and Asian mothers (95.4%).
Conclusions
In our general obstetric population, birthweight customization identified more SGA infants at risk of perinatal mortality and morbidity compared with the INTERGROWTH-21st standard. The INTERGROWTH-21st standard failed to detect many at-risk SGA infants, particularly among ethnic groups with larger maternal size while disproportionately identifying higher rates of SGA among those with smaller maternal size. Local validation is needed prior to implementation of the INTERGROWTH-21st standard to avoid misclassification of infant birth size.
Recent publication of the multinational INTERGROWTH-21st Project standards for newborn anthropometry has provided a new benchmark for international comparisons across multiethnic populations. Importantly, in women at low risk of fetal growth impairment, optimum infant size at birth was described as almost identical among the 8 included countries. This international standard is intended for use in clinical practice both within populations and for comparisons between nationalities. The ability of the INTERGROWTH-21st standard to identify infants at risk of adverse outcomes has not yet been reported.
Population birthweight standards have traditionally been used to identify infants who are small for gestational age (SGA) that may have experienced intrauterine growth restriction. Such infants are at increased risk of neonatal death and morbidity; however, use of population birthweight standards may underestimate risk for some infants and overestimate risk for others. For example, preterm birth is inherently pathological and population standards consistently underestimate optimal birthweight in preterm infants compared with ultrasound estimates of fetal weight at preterm gestations in infants subsequently born at term. Conversely, in ethnic groups with smaller-than-average maternal size, some SGA infants are constitutionally small and not growth restricted and vice versa.
Customized birthweight standards differ from population standards in that they use ultrasound-based measures of fetal size and account for maternal characteristics that influence birthweight, including maternal height, weight, parity, and ethnicity. Infants who are SGA by customized criteria generally show increased rates of perinatal morbidity and mortality compared with population birthweight reference data.
New Zealand has a multiethnic birthing population with more than a third of women of Maori or Pacific Island ethnicity, groups that were not included in the INTERGROWTH-21st standard. Our aims were to compare the following: (1) the INTERGROWTH-21st population birthweight standard to a customized birthweight standard in a general obstetric population from Auckland, New Zealand, for the detection of SGA infants at increased risk of neonatal mortality and morbidity, and (2) rates of stillbirth among infants classified as SGA between the respective standards.
Based on previous work, we hypothesized the following: (1) infants classified as SGA by customized criteria alone compared with INTERGROWTH-21st criteria alone would have a higher risk of neonatal mortality and morbidity, and stillbirth; (2) infants classified as SGA by INTERGROWTH-21st criteria alone would have similar neonatal outcomes to infants not SGA by either standard; (3) obstetric risk factors for SGA would be more common in mothers whose infants were classified as SGA by customized criteria alone compared with INTERGROWTH-21st criteria alone; and (4) compared with those of European ethnicity, the incidence of SGA by the INTERGROWTH-21st criteria would be lower in those of Pacific Island ethnicity and higher in those of Indian and Asian ethnicities.
Materials and Methods
This was an analysis of prospectively gathered maternity data from National Women’s Health (NWH), Auckland City Hospital, Auckland, New Zealand, from January 2006 to December 2013. NWH is a tertiary referral hospital with a diverse ethnic population and approximately 7500 births annually. The NWH database records maternity data for all births occurring ≥ 20 weeks’ gestation, including demographics, antenatal complications, delivery details, and neonatal outcomes. Data are routinely checked for completeness, outliers, and inconsistency. Ethical approval was obtained from the Research Board of the Auckland District Health Board (study number 4632).
To accord with data available from the INTERGROWTH-21st project singleton infants born at NWH from 33 to 42 weeks’ gestation without major malformations were included in this analysis. Infants were excluded if mothers were unbooked, transferred to NWH during pregnancy or labor, or if data for maternal customization were missing.
Gestation-specific customized birthweight centiles were calculated as previously described using locally derived coefficients, adjusting for maternal height and weight at booking, parity, ethnicity, and infant sex. Birthweights were compared with both the customized and INTERGROWTH-21st birthweight centiles, with SGA defined as < 10th centile. Infants were categorized as follows: SGA by both criteria (SGA-both), SGA by INTERGROWTH-21st criteria alone (SGA-IG only), SGA by customized criteria only (SGA-cust only), or not SGA by either criterion (non-SGA).
Gestational age was calculated from the first day of the last menstrual period (LMP) if certain, adjusted if fetal ultrasound measurements differed from LMP gestational age according to the Australasian Society for Ultrasound in Medicine guidelines or by dating ultrasound if the LMP was uncertain. The majority of women at NWH undertake a first-trimester ultrasound. From 2009, gestational age for stillborn infants was defined as the estimated gestation at time of death, determined by maternal report of when fetal movements ceased. We used the median latency between estimated gestation at death and birth of these stillbirths (2 days) to estimate gestation at death for stillbirths from 2006–2008.
Maternal height (centimeters) and weight (kilograms) were measured at the first antenatal visit. Parity was defined as the number of liveborn infants of any birthweight or gestation or stillborn infants from 20 weeks’ gestation or where the infant weighed 400 g or more if gestation was unknown. Self-reported maternal ethnicity was grouped and prioritized in order of Maori, Pacific Peoples, Asian, Indian, Other and European. Asian ethnicity included women from China, South-East Asia, Japan and Korea. Indian ethnicity included women from India and those of Fijian-Indian origin.
Risk factors for SGA were smoking during pregnancy and pregnancy-induced hypertension (PIH), defined as gestational hypertension or preeclampsia/eclampsia. Preterm birth was defined as delivery at less than 37 weeks’ gestation.
Groups were compared for the primary outcomes of the following: (1) composite adverse neonatal outcome, defined as neonatal death (death occurring within the first 28 days of life of a liveborn infant) or neonatal morbidity, defined as admission to neonatal intensive care unit (NICU) for > 48 hours, positive pressure respiratory support > 4 hours or 5-minute Apgar score < 7; and (2) stillbirth. These neonatal morbidity measures have been shown to be important predictors of adverse neonatal outcome.
Statistics
Statistical analysis was performed using SAS 9.4 (SAS Institute Inc, Cary, NC). SGA and ethnic groups were compared using generalised linear models with adjustment for multiple comparisons. Risk ratios for primary outcomes were estimated by Poisson regression with robust error estimates. Primary outcomes are presented as risk ratios with 95% confidence intervals using the non-SGA group as the referent. An adjusted two-sided value of P < .05 was considered statistically significant. Stillborn infants were excluded from analyses for neonatal outcomes. A sensitivity analysis of primary outcomes confined to term infants was performed, using the same methodology as above.
Sensitivity, specificity, and predictive values were calculated for composite adverse neonatal outcome for both birthweight standards.
Results
A total of 56,638 singleton nonanomalous infants were born at 33–42 weeks’ gestation from January 2006 to December 2013. Customized birthweight centiles were calculated for 53,484 infants because 3152 women had missing data for customized centiles (maternal height or weight, n = 3150, and birthweight, n = 2). Nearly half of mothers were of European ethnicity, followed by Asian (20.3%), Pacific (13.5%), Indian (7.4%), Maori (6.8%), and other ethnicity (3.5%) ( Table 1 ). Mean (SD) maternal body mass index (BMI) at booking was 25.3 (6.1) kg/m 2 and gestational age at delivery was 39.4 (1.5) weeks ( Table 1 ). Preterm birth occurred in 5.1% of our cohort ( Table 1 ).
| SGA classification | Total (n = 53,484) | Non-SGA (n = 47,090) (88.0%) | SGA-IG only (n = 172) (0.4%) | SGA-cust only (n = 4015) (7.5%) | SGA-both (n = 2207) (4.1%) | P value |
|---|---|---|---|---|---|---|
| Maternal characteristics | ||||||
| Age, y | 31.4 (5.6) | 31.4 (5.6) a | 28.7 (4.5) b | 31.4 (5.8) a | 31.0 (5.8) c | < .05 |
| Nulliparous | 25,857 (48.4%) | 22,552 (48.0%) a | 154 (89.5%) b | 1698 (42.5%) c | 1399 (64.1%) d | < .05 |
| Ethnicity | a | b | a | b | ||
| Maori | 3659 (6.8%) | 3128 (6.6%) | 0 (0.0%) | 372 (9.3%) | 159 (7.2%) | < .05 |
| Pacific | 7248 (13.5%) | 6272 (13.3%) | 1 (0.6%) | 768 (19.1%) | 207 (9.4%) | |
| Asian | 10,829 (20.3%) | 9658 (20.5%) | 55 (32.0%) | 546 (13.6%) | 570 (25.8%) | |
| Indian | 3936 (7.4%) | 3331 (7.1%) | 109 (63.4%) | 100 (2.5%) | 396 (17.9%) | |
| Other | 1867 (3.5%) | 1649 (3.5%) | 7 (4.0%) | 121 (3.0%) | 90 (4.1%) | |
| European | 25,945 (48.5%) | 23,052 (49.0%) | 0 (0.0%) | 2108 (52.5%) | 785 (35.6%) | |
| Height | 165 (6.8) | 165 (6.8) a | 154 (5.2) b | 166 (6.3) c | 162 (6.7) d | < .05 |
| Weight | 69.3 (17.9) | 69.1 (17.7) a | 48.7 (5.9) b | 75.6 (19.8) c | 63.7 (15.3) d | < .05 |
| BMI, kg/m 2 | 25.3 (6.1) | 25.3 (6.0) a | 20.3 (2.6) b | 27.2 (7.0) c | 24.2 (5.4) d | < .05 |
| SGA risk factors | ||||||
| Smoking | 4100 (7.7%) | 3253 (6.9%) a | 1 (0.6%) a | 581 (14.5%) b | 265 (12.0%) c | < .05 |
| PIH | 3234 (6.1%) | 2548 (5.4%) a | 8 (4.7%) a,b | 386 (9.6%) b | 292 (13.2%) c | < .05 |
| Infant characteristics | ||||||
| Gestation, wks | 39.4 (1.5) | 39.5 (1.4) a | 39.3 (1.2) a,b | 39.3 (1.7) b | 38.8 (1.8) c | < .05 |
| Preterm birth | 2745 (5.1%) | 2131 (4.5%) a | 6 (3.5%) a,b | 331 (8.2%) b | 277 (12.5%) c | < .05 |
| Birthweight, g | 3433 (515) | 3526 (457) a | 2667 (195) b | 2893 (317) c | 2469 (349) d | < .05 |
| Birthweight term infants, g e | 3475 (478) | 3560 (428) a | 2684 (175) b | 2957 (233) c | 2562 (238) d | < .05 |
| Neonatal death f | 12 0.2/1000 | 9 a 0.2/1000 | 0 0.0/1000 | 0 0.0/1000 | 3 b 1.4/1000 | < .05 |
| NICU admission > 48 h f | 2032 (3.8%) | 1390 (3.0%) a | 6 (3.5%) a,b | 283 (7.1%) b | 353 (16.2%) c | < .05 |
| Ventilation > 4 h f | 1207 (2.3%) | 975 (2.1%) a | 3 (1.7%) a,b | 136 (3.4%) b | 93 (4.3%) b | < .05 |
| Apgar score < 7 at 5 min f | 651 (1.2%) | 457 (1.0%) a | 1 (0.6%) a,b | 60 (1.5%) b | 36 (1.7%) b | < .05 |
| Composite neonatal outcome f | 2695 (5.0%) | 1974 (4.2%) a | 6 (3.5%) a,b | 338 (8.4%) b | 379 (17.4%) c | < .05 |
| Stillbirth | 97 1.8/1000 | 59 a 1.3/1000 | 0 0.0/1000 | 15 b 3.7/1000 | 23 b 10.4/1000 | < .05 |
a-d Indicate different groups ( P < .05 with adjustment for multiple comparisons)
e Infants > 37 weeks’ gestation: n = 50,739 (non-SGA, n = 44,959; SGA-IG only, n = 166; SGA-cust only, n = 3684; SGA-both, n = 1930)
f Denominators for neonatal outcomes exclude stillbirths: n = 53,387.
The incidence of SGA was 4.5% and 11.6% by INTERGROWTH-21st and customized criteria, respectively. Of infants identified as SGA by INTERGROWTH-21st criteria, 89% were also identified as SGA by customized criteria. The customized standard, however, identified an additional 7.5% of infants as SGA ( Table 1 ).
The incidence of neonatal death or morbidity was 17.2% in infants identified as SGA by INTERGROWTH-21st and 12.1% in those identified as SGA by customized criteria. The risk of this composite adverse neonatal outcome was highest in the SGA-both group (RR, 4.1, 95% CI, 3.7–4.6). Composite adverse neonatal outcome was also twice as common in SGA-cust only infants (RR, 2.0, 95% CI, 1.8–2.2), but risks were not increased in infants identified as SGA-IG only (RR, 0.8, 95% CI, 0.4–1.8) ( Figure ). Similarly, the risk of stillbirth was highest in the SGA-both group (RR, 8.3, 95% CI, 5.1–13.4) and was also increased 3-fold in SGA-cust only infants (RR, 3.0, 95% CI, 1.7–5.3). There were no stillbirths among infants who were identified as SGA-IG only ( Figure ).
In a sensitivity analysis that excluded infants born at less than 37 weeks’ gestation, results were very similar (n = 50,739); the risk of adverse neonatal outcome was increased in the SGA-both (RR, 3.3, 95% CI, 2.8–3.8) and SGA-cust only (RR, 1.5, 95% CI, 1.3–1.8) groups but not in those classified as SGA-IG only (RR, 1.1, 95% CI, 0.5–2.6). In term-born infants, the risk of stillbirth was highest in the SGA-both group (RR, 6.0, 95% CI, 3.1–11.5) and was also increased in the SGA-cust only infants (RR, 2.6, 95% CI, 1.2–5.2).
The sensitivity and specificity of INTERGROWTH-21st criteria for composite adverse neonatal outcome among SGA infants were 14.3% (95% CI, 13.0–15.7%) and 96.1% (95% CI, 95.9–96.3%), respectively, with positive and negative predictive values of 16.3% (95% CI, 14.9–17.9%) and 95.5% (95% CI, 95.3–95.7%), respectively. For customized criteria, sensitivity and specificity were 26.6% (95% CI, 24.9–28.3%) and 89.2% (95% CI, 88.9–89.5%), respectively, with positive and negative predictive values of 11.6% (95% CI, 10.8%–12.4%) and 95.8% (95% CI, 95.6–96.0%).
Compared with the non-SGA group, obstetric risk factors for SGA (smoking and PIH) were approximately twice as common in SGA-cust only and SGA-both groups ( P < .05) but were not more common in the SGA-IG only group ( Table 1 ). Preterm birth was also more common in the SGA-cust only and SGA-both groups ( P < .05) but not in the SGA-IG only group ( Table 1 ).
INTERGROWTH-21st birthweight Z-scores ranged from a mean (SD) of –0.2 (1.0) in infants born to Indian women to 0.8 (1.1) in those born to Pacific women ( Table 2 ). Compared with European, infants of Indian mothers were more than 4 times as likely to be classified as SGA by INTERGROWTH-21st criteria, and for Asian infants, twice as likely ( P < .05, Table 2 ). However, the rate of SGA by INTERGROWTH-21st criteria for Maori, Pacific, and European infants was less than a third that of customized criteria, and for Asian and infants in the category of other, this ratio was approximately half.
